(414g) Evaluation of Real Time Optimization – Model Predictive Control and Economic Model Predictive Control Strategies for Post Combustion CCS Operations within Refineries

Carbon Capture and Storage (CCS) technologies has become key in abating post combustion CO₂ emissions from large point sources, including powerplants, petroleum refineries, and chemical production facilities. Although, CCS technology has seen an abundance of scientific efforts being invested in capturing CO₂ emissions from power plants, lately, this technology is seeing an increased interest within the petroleum refinery sector as a method to decrease carbon dioxide emissions. According to available data, ~ 65 % of the CO₂ emissions from refineries are due to stationary combustion of natural and fuel gases. Post combustion CO₂ capture, therefore, is vital for the overall CO₂ emissions reduction within refineries.

Captured CO₂ will most likely be transported over long distances via pipelines. CCS operations, therefore, can be considered to consist of three energy intensive sub-operations: CO₂ capture, CO₂ compression, and CO₂ dehydration to meet pipeline specifications. A steady state simulation of a CO₂ capture process for a 300,000 bbl/day refinery shows that ~150 MW of energy is required to regenerate CO₂ from the amine regeneration unit to capture 90 % of the available post combustion CO₂ emissions. Additionally, ~ 14 MW of energy is required to compress and dehydrate the CO₂ to pipeline conditions. Hence, the energy intensive nature of the three CCS sub-operations requires one to consider them as a combined system. Additionally, due to the inherently dynamic refinery processes, the CCS operations rarely reach steady state. Hence, the CCS operation must be studied in a transient state for better economic optimization and control. In this study, a dynamic optimization and control strategy is developed for a CCS system in context of a standard mid-sized refining facility.

To begin, steady state simulation of the CCS operation was performed using an aqueous solution of 30 wt % monoethanolamine (MEA) to capture CO₂ and a 98 wt % triethylene glycol (TEG)/water to dehydrate the captured CO₂ gas. These absorbents were chosen due to current industry practices for CO₂ capture and dehydration technologies. A 7-stage CO₂ compression unit was used to compress the gas to pipeline conditions (15.17 MPa). For a refinery of this size, results indicated that at least two absorber units, one CO₂ regeneration unit, one dehydration, and one glycol regeneration unit would be required. Two design limits were imposed on the simulation while determining the number of columns: Height to Diameter Ratio (H/D) > 2 and Diameter of Column (D) < 12 m. A minimum threshold of 90 % CO₂ recovery was maintained. The steady state layer was then optimized using a cost function. This layer was then linked to the Model Predictive Control layer, thereby forming a Real Time Optimization – Model Predictive Control (RTO-MPC) optimization and control infrastructure for the CCS operation. The RTO-MPC infrastructure is considered as an industry standard while implementing optimization and control processes. An important assumption of the RTO-MPC infrastructure is that the process eventually reaches steady state. However, this assumption is not likely to occur for CCS capture operations within refineries. Therefore, the result of the RTO–MPC layer may not be an accurate representation of a real time process control operation with respect to the process economic performance.

To overcome this problem, a novel Economic Model Predictive Control (EMPC) strategy was developed in which the process cost function was optimized in a dynamic and transient manner with respect to the process constraints in a single layer. Unlike the RTO-MPC process, EMPC does not require the process to be optimized at steady state conditions. In addition, the EMPC strategy uses a single layer to optimize and control the cost function. The hypothesis is that such a control strategy will be able to reduce the response time to optimize the cost function, thereby achieving a better economic optimization of the entire CCS process. This study compared the economic performance of the two mentioned control strategies for CCS operation within the refinery.

Once the base case (using MEA and TEG absorbents) was completed, the study was expanded to replace MEA and TEG with ionic liquids, Trihexyl(tetradecyl)phosphonium 2-cyanopyrrolide ([P₆₆₆₁₄][2-CNPyr]) and 1-Ethyl-3-methylimidazolium methanesulf onate [EMIM][MeSO₃], respectively. According to the literature, [P₆₆₆₁₄][2-CNPyr] has shown to possess high CO₂ absorption capacity, low regeneration energy, and negligible solvent loss during regeneration. Additionally, [P₆₆₆₁₄][2-CNPyr] has also shown to be less sensitive to H₂O on the ionic liquid due to the anion [2-CNPyr]^-. [EMIM][MeSO₃] has shown to be highly hygroscopic while also possessing the benefits of low regeneration energy and negligible solvent losses and was studied as a replacement for TEG.